Jump to content
IGNORED

RGB2VGA: customizing a GBS-8200 for a cleaner low-lag image


mytek

Recommended Posts

18 hours ago, gozar said:

That's weird, my 2007FP looks really good, whether I'm using the SVideo or my GBS8220:

It does indeed ? . Maybe mine's on its last legs, and/or needs a capacitor change, because it looks poor with S-Video input as well. So we won't rule that one out based on my poorly working unit :) .

 

  • Like 1
Link to comment
Share on other sites

OK so I have it installed and I put it through its paces today.

 

Both 50hz and 60hz display ok on my monitor, the aspect ratio is just different. My 1088XEL is a PAL so the GBS programmer needed to be using the PAL modes to all fit on the screen.


Switches like this....

 IMG_2938.thumb.jpeg.f5e51ae3b2eccf9c0533dd0bd62e2efb.jpegIMG_2935.thumb.jpeg.2bef3b51f096468c57a46bf9e6d91a7f.jpegIMG_2937.thumb.jpeg.de51ff29ad4ddca6cae4c7767fbc089d.jpeg

 

And like this...


IMG_2939.thumb.jpeg.f8911d06b83d3b46ed4d7b104973000d.jpegIMG_2940.thumb.jpeg.6c5484e31f0c5e12e92039b5959939b2.jpegIMG_2936.thumb.jpeg.2d5b62b5b2a2da5ec862610016369b5b.jpeg

IMG_2941.thumb.jpeg.81d50501d8111ecb561f1cc8c923dbea.jpegIMG_2942.thumb.jpeg.e9a9dc9c2165d7b22a276f73bfb15a2f.jpeg

 

This image is visibly flickery, it's 480i so it's always flickery, I thought it was going to get deinterlaced and I would get a solid display. Tried all four combinations of dip switch settings, it was always flickery. 

IMG_2932.thumb.jpeg.ac20e2c6dc22dedd0737b4ecd3c46c47.jpeg

 

This image is interesting, with the stock firmware the GBS would output something my monitor couldn't handle and I would just get a no supported screenmode message on the screen. With the custom firmware I get this... 

IMG_2922.thumb.jpeg.2d11b22267a8692ebc93931d0e66ee81.jpeg

Win!!

 

Here are some more bits of candy and closeups.

IMG_2915.thumb.jpeg.a8d2dbdf86babad16f6e26bad8baae2a.jpegIMG_2902.thumb.jpeg.87a7e2a9a993c54a32d78c49f869b043.jpegIMG_2930.thumb.jpeg.036ae8001ff1811bdef385f39321dbb9.jpeg

 

@Stephen I've made up your kit and I'll be posting it tomorrow while I'm out

 

IMG_2940.jpeg

  • Like 5
Link to comment
Share on other sites

On 10/25/2019 at 11:28 PM, mytek said:

The stock GBS will not necessarily be set to use the RGBS input by default. You would have to get into it's menu and select the correct input, as well as pick an output resolution. And then there are a multitude of other adjustments. Or better yet wait until you have the GBS-PGM board and enjoy the simplicity :) .

 

The GBS will need CSYNC connected via a 470 ohm resistor to J27 on the XEL as well.

 

*grumble* *grumble* *GRUMBLE*, after I hooked up the Csync, I finally started getting a picture but the blue channel, and very dim.  Put a scope on the different channels, only blue was showing a signal.  I was thinking crap, the VBXE is busted.  At that point for some reason the board wouldn't draw power from the 1088, I was worried something got blown on it.  Started checking the power/ground planes, everything looked to be still connected, but still wouldn't power, was also checking VBXE schematic to check traces.   I finally checked the power harness wire, and + had lost continuity.  Which then it dawned to me, maybe that crappy wiring harness they sent with the GBS-8200 has issues.  Sure enough, some of the wires were not conducting.  *GRUMBLE*

 

Re-built a new wiring harness, and low and behold, I get a picture, but still having some issues, now the firmware on this damn thing doesn't want to respond to the buttons, when it was just hunky dory before, sometimes the picture comes up, sometimes not, so instead of beating my head against the wall some more, I will wait to get my RGB2VGA boards, and hopefully that will solve all problems.  I'm not completely convinced the wiring harness isn't still to blame, but I'm not gonna touch it until I get the boards.

Link to comment
Share on other sites

Mr Robot looks like you need to fiddle with the pitch setting on the monitor setup (maybe decrease its value) to get more of the image on screen, because it looks a bit wide right now, even when using the best fitting mode. After doing that you'll likely need to use the Hcenter setting to get things centered up again.

 

Otherwise it's looking very nice ? .

 

With those presets there should be no de-interlacing occurring, but I notice that some of the custom graphics modes such as FlickerTerm don't work correctly. I don't have an explanation as to why, but now that the code has been brought into the Basic dialect, and is well commented perhaps someone will do some experimenting to get this aspect working. But truthfully there isn't much other than a few demos and maybe one or two games that take advantage of this trickery to gain additional resolution.

 

Edit: I believe de-interlace was deactivated due to it causing a soft somewhat blurry image, and in the original GBS firmware didn't correct the problem you are seeing.

 

wildstar87 sounds like you are getting closer. Keep in mind that the original firmware in the GBS was always problematic with the Atari, hence the reason we are taking this alternative approach instead. Nice thing is the 2 dip switches allow for the 4 different modes that should work right out of the box for either the Sophia or VBXE, so not much fiddling required.

 

Link to comment
Share on other sites

Testing RGB2VGA-XLD prototype board from JLCPCB (via EasyEDA). Came to a total of $27 for 5 boards including DHL shipping. Total time from order to landing on my door step was 7 days.

 

Testing in process...

RGB2VGA-XLD_running.thumb.JPG.d4941fa0e6ce7fa9166f14ca65b2a58f.JPG

 

Shown Installed in 1088XLD

RGB2VGA-XLD_proto_installed.thumb.JPG.bc310265e4a3664255879bf4c99e8e76.JPG

 

Complete Assembly

RGB2VGA-XLD_top_complete.thumb.JPG.dc69b074ab148e8a6e0e7604a65e8a38.JPG

 

Top View of Carrier Board Only (board has double ground plane)

RGB2VGA-XLD_top.thumb.JPG.36e490b81436ccfeda42d7198e3c9aed.JPG

 

Bottom View of Carrier Board Only

RGB2VGA-XLD_bot.thumb.JPG.a996552619bc001b0359349a5f15fbaf.JPG

 

During testing, and as was expected, the 74HC4053 triple analog switch 'ON' resistance required a boost to the output of the Sophia's ADV7125 DAC chip, in order to restore the 0.7V RGB signal. This is relatively easy to accomplish by adding an additional 360 ohm SMD resistor on top of the existing 561 ohm RSET resistor, or in my case since I kind of butchered a trace when removing that resistor in the past, I opted to solder in an 1/8 watt 220 ohm thru-hole resistor instead.

 

If you click on the image to enlarge, you'll see the ripped off pad and trace where the original SMD RSET resistor went.

RGB2VGA-XLD_Sophia_220ohm_RSET_resistor.thumb.JPG.67091f93199217ee516bbc155c5c1e13.JPG

 

Not sure what can be done on the VBXE to boost its output signal or gain. Hopefully its DAC has a provision for setting this as well. Anyone know off hand which DAC chip it uses? (see EDIT below)

 

EDIT I found this file on Candle's website that covers technical details of all the VBXE versions: release-install.zip

          Between the various versions of the VBXE, two different dedicated Video DAQ chips were employed,

          which ended up being replaced by resistor DACs followed by buffer amplifiers in the current version.

          So no standard method of adjusting output voltage would equally apply.

 

The end result is that I get a clean 0.7V output from either the RGB or the VGA, when switching between the two. And the switching is accomplished by either using a shorting jumper on the RGB2VGA-XLD board's J9 header, or wiring an SPST external switch to this header. With no shorting jumper or switch installed the default is to produce a VGA output.

 

I've been making some minor changes to the GBSCTRL firmware, brought about by changing from PAL chips over to NTSC, which revealed what I came to find out is a "tear line". This is an invisible line that travels either up or down the screen causing a slight offset of the image to occur. And although it's subtle, if it's moving fast (which it was) you tend to notice it.

 

Here's a quick explanation provided by rama over at the shmups forum...

Quote

It's where start and end of the buffer meet, causing the 2 points to represent a different point in time.

Link to that conversation

 

I was able to tweak a preset to slow it's screen travel wayyyyy down, which has the benefit of it no longer being an eye catcher.

 

Didn't see the effect with PAL chips. But I haven't gone back to that setup to scrutinize it yet.

 

While I was tweaking the GBSCTRL code, I decided to add a method to identify versions. So what I did was add two constants to the top of the code listing called Major and Minor. By changing these to match the current revision level (i.e., V1.2 breaks down as Major = 1 and Minor = 2), these will be used to flash the GRN LED. So for V1.2 you would see a single long flash followed by two short flashes. And after a brief pause this will be repeated. RED will still denote "BUSY" meaning that new presets are being loaded.

 

Edit: I'll be simplifying the Version ID to work off of a single constant called "Version" which you can simple equate to 12 to represent V1.2.

 

 

Edited by mytek
Added VBXE technical data and notes
  • Like 3
Link to comment
Share on other sites

Next piece to the puzzle ready to go to fabricators.

 

VGA2GBS Adapter Board (TOP VIEW)

 

VGA2GBS_top.thumb.png.b0bba01173ef9f50fa5f5a9dbf1bdcbe.png

 

 

VGA2GBS Adapter Board (BOTTOM VIEW)

 

VGA2GBS_bot.thumb.png.d1dd365d94fcd569506d914907727498.png

 

This will allow a standard high density VGA connector to mount in place of one of the 1088XLD's SIO Jacks, of which there are two. Since it's designed to interface to the the GBS's JST VGA connector, it'll also work for a non-1088XLD application with either the GBS-8200 or GBS-8220 where you might wish to mount it internally and require a way to get the VGA to the outside world.

 

Will be using this 2mm JST cable to get from GBS board over to the adapter board.

 

JST-XH_8-pin_cable.png.fe5cac75553ec52c9838881901d6df85.png

Ebay Buy Link

Datasheet

 

The provided PCB mount connector won't be used, and instead the loose wire end will be trimmed to an appropriate length and terminated to a standard 8-pin Dupont style female connector.

 

  • Like 3
Link to comment
Share on other sites

Due to problems with the video switching circuit I am using on the RGB2VGA-XLD board, there will be yet another revision before the design will be released.

 

So for those that don't understand the problem, or what the video switching is actually being used for, here's an explanation...

 

On the 1088XLD the RGB video coming out of either a Sophia RevB or a VBXE video board is routed out to a DIN-13 jack using a pin-out similar to an Atari ST. So in order to have VGA output when installing the GBS-8200 RGB to VGA converter, it meant giving up the possibility of having the normally provided RGB output. This was due to not being able to feed two RGB devices from a single RGB output without encountering excessive loading of the RGB source, thus rendering a very dim image on both the RGB and the VGA monitors.

 

The solution was to either switch the RGB, and only feed it to one device at a time, or buffer the RGB going to both devices. I chose switching of the RGB thinking that would be the easiest and cheapest to implement. Problem is that the electronic triple analog switch I used has an inefficiency in it's switches which end up looking like something between 80-100 ohms of resistance when run from a 3.3V supply. This wouldn't be a problem in a high impedance application, but it proved problematic due to the low impedance of RGB. This resulted in a signal with less amplitude and hence a dimmer picture. In order to counteract this, I increased the output drive of the Sophia by adjusting a single gain setting resistor. This worked great for the Sophia, but it won't work for the latest VBXE which is based on an entirely different method of digital-to-analog conversion.

 

So now I'm left with a decision, and that is to implement buffering of the RGB video. There are two approaches I can take in this regard, one being to use a buffer with additional gain and still retain the electronic switching. So in essence move the amplitude increase to a separate device, thus leaving the Sophia or VBXE stock. And the 2nd approach would be to use two 1:1 RGB buffer chips individually feeding the GBS and the DIN-13 jack, thus eliminating the switching, and also allowing simultaneous use of both the RGB and VGA outputs. The 2nd approach is the better one.

 

Next comes the cost, which I am trying to keep very reasonable, so if I'm to use two buffer chips they need to be inexpensive and readily available. So in my searching I came across the THS7314 as the device that fits the bill, which can be purchased for only $0.63 from Mouser. However its not a thru-hole component, so it will be a bit more challenging from a DIY perspective. But it is one of the larger SOIC-8 SMD packages, so not too terrible for hand soldering. It's also available from multiple sources and shows no signs of going obsolete anytime soon. And If I had known about this device in the beginning of my journey, I would have never fussed around with video switching in the first place ;) .

 

So stay tuned for yet another prototype layout, build, and testing.

 

  • Like 3
Link to comment
Share on other sites

New PCB layout for the RGB2VGA-XLD incorporating changes as talked about in previous post. Now sporting dual RGB video buffer chips for complete isolation between RGB and VGA outputs.

 

TOP

RGB2VGA-XLD_top.thumb.png.79fae95d1ec189ab4ca451c2f75ec345.png

 

BOTTOM

RGB2VGA-XLD_bot.thumb.png.b84eae9f259921accc96ab0fe0990451.png

 

Schematic: RGB2VGA-XLD_v1_1_schema.pdf

 

Edited by mytek
Added Schematic
  • Like 3
Link to comment
Share on other sites

Well finally got a picture up, which makes me breathe a bit easier, though the color is a bit green, and I am seeing noise, but haven't made the other suggested mods to the GBS yet, hopefully that will fix things.  Thanks Mytek for this board, otherwise it wouldn't be working at all.

Edited by wildstar87
Link to comment
Share on other sites

2 hours ago, wildstar87 said:

Well finally got a picture up, which makes me breathe a bit easier, though the color is a bit green, and I am seeing noise, but haven't made the other suggested mods to the GBS yet, hopefully that will fix things.  Thanks Mytek for this board, otherwise it wouldn't be working at all.

If you try the copper foil on the bottom of the GBS board and/or the capacitor change let us know if you see an improvement.

 

However I am a bit surprised that you are seeing much noise. I only see a tiny bit when looking at something with a very dark background such as the dark green when TBXL boots in NTSC. Also my color looks well balanced. I double checked the drive on the RGB in my monitor and they are all set to identical parameters. Perhaps the trim pots on the RGBS inputs of the GBS are not all the same.

 

Link to comment
Share on other sites

11 hours ago, Mr Robot said:

I've noticed with the XEL board that if I leave the XEL off but the GBS on and the screen on, eventually I start to get all sorts of odd displays. Next time I see it I'll post pics.

Not totally surprised. Ever think of running the GBS from the switched 5V available from one of the 3 auxiliary power headers?

 

Link to comment
Share on other sites

Ok, so I did some of the mods, replaced C11 with 22uF, removed the pots and direct connected the inputs, copper tape, 75 ohm impedance mods, didn't have the parts for the other mods.  A slight improvement, but nothing drastic.  Still seems a bit on the green side.  In Star Raiders when I turn the shields on, they are green, not blue like on the S-video CRT.  Blue with greenish hue in Dos, not blue like CRT, or the pictures above.  I get diagonal waves going across the screen that are periodic.  I haven't tried a different LCD yet, but based on what the monitor looked like with Sophia, not expecting much of a difference.  I haven't tried any of the VBXE specific stuff yet, just standard output so far.

Interesting to note though, after the mods, the normal firmware is working now, where as before, it was completely non-functional.  I may try and tinker with some of the parameters in the firmware, or I may just try to mod the GBS with the other developed firmware, though not crazy about using a web interface to adjust things.

Link to comment
Share on other sites

Moving diagonal lines that randomly appear and disappear is purely related to the quality of the power source being used to supply the GBS board. If the power supply is a switching type, which many of the Chinese wall warts are, you are likely seeing the result of poor filtering in that power adapter allowing the switching pulses to leak through and get into the video circuits. Normally this is more of a problem with using a 12VDC supply than a 5VDC one for the GBS. With the higher voltage input driving the onboard 3.3V regulator, it has a more difficult time filtering out noise derived from the power source.

 

When I was developing the RGB2VGA-XLD version for the 1088XLD, I ran into this exact same problem due to wanting to use the 12VDC available on that motherboard which was not as well filtered as the 5V supply. Thus the reason an additional filtering circuit was added to the power input side of the RGB2VGA-XLD board, which completely cured the diagonal line issues. Check out the recently posted schematics to see what that filtering circuit looks like.

 

Unfortunately if I had known that this was the noise you spoke of before doing all the mods, I would have mentioned trying a different power source for your GBS instead. Since my experiments and development are mainly centered around creating a VGA output for the 1088XLD, the noise issues were fixed back when I first began the experiments and are no longer present. Out of sight, out of mind :) .

 

Edit: not sure if the power supply could contribute to the greenish tint you are seeing. But if it is, then perhaps that will go away as well.

 

 

 

Link to comment
Share on other sites

I'm taking the +5V off the 1088XEL aux header, but the 1088XEL is being powered by the CUI +5v PS that I believe was suggested for the 1088XEL.  The RGB2VGA-XLD board that has the filtering, I'm assuming that is the big carrier board, not the small add-on board.  I'll take a look at the schematic for the filtering, can I assume that it can be used on the +5v line as well?  As for the green tint, unfortunately I have no way of seeing if the output from the VBXE is color correct, looking at more VBXE threads some people were complaining about the NTSC output, since it is using PAL palette?  I don't know if it's the GBS or VBXE that is doing it, I may have to put the pots back on the board, but I'm dubious at adjusting using those, as the signal is weak as is.  I'm wondering if I should just get an OSSC, but of course that is a lot more $$$.

 

My display looks very much like the one pictured here in Dos. 

 

Edited by wildstar87
Link to comment
Share on other sites

If you are using the 5V AUX power on the XEL, there is already additional filtering present.

 

Wow not sure what's going on with your setup. I thought for sure it was a power source problem based on your description of the noise (scratching head). I don't have a VBXE to test with in order to rule that out (only a Sophia over here). I think Mr Robot has been using a VBXE with his GBS, and I haven't heard of any noise issues from his side.

 

Link to comment
Share on other sites

That's funny, because my UAV picture on a Commodore 1702 is stunning, it's what I'm comparing the VBXE output to LCD with.  I can definitely see the difference between the blending that is inherent in NTSC CRT, vs the obvious sharp but jagged lettering on LCD.  The greenish hue to color is what concerns me the most, and then the noise.  I'm wondering if I just got a bad board.  I also got the tear line bar, most noticeable when in 1088XEL bios menu.  I have a ESP8266 module on the way to try out the GBSControl firmware, may have adjustments that will fix issues.

Edited by wildstar87
Link to comment
Share on other sites

RGB2VGA-XLD Final V1.1 Board Version Completed and Testing

It took three versions to arrive at V1.1, with the first one being an ATTiny MCU based prototype with dedicated VGA only operation, having two resolution settings.

 

By the time I got to V1.1, four independent resolutions modes were selectable, full RGB buffering had been added, thus allowing simultaneous operation of an RGB, as well as a VGA monitor, with neither interfering with the other. With the implemented buffering, it will no longer require any modification of either a Sophia RGB or VBXE board's gain. They should work right out of the box.

 

And to make for a more seamless look, a separate VGA connector board was created that mounts in the 2nd redundant SIO jack position.

 

Because of the PIC MCU, flashing the firmware can be done with a JOY2PIC using your Atari 8-bit computer.

 

Assembled GBS Carrier/Mounting Board

RGB2VGA-XLD_V1_1_boardtop.thumb.jpg.8e0b6e11f7d50e091d794588d71cca9c.jpg

 

 

Complete Unit (GBS-8200 mated to Carrier Board)

RGB2VGA-XLD_V1_1_complete.thumb.JPG.0022bfb5460bde177f49c81530b41719.JPG

 

VGA2GBS Connector Mounting Board Assembled

VGA2GBS_front.thumb.JPG.bdc76cf234a1f0cf2f8b9934bfcfea6e.JPG

VGA2GBS_rear.thumb.JPG.4529c41117ae215dc152daeef81d5331.JPG

 

Bottom Protrusions Sanded Down for Clearance

VGA2GBS_bottom.thumb.JPG.35b489c4598e3ac35e79c38145784d53.JPG

 

1088XLD Motherboard Prepped for VGA Connector Board Install (#6 Nylon Washers with 4-40 hardware)

Prep_for_Install.thumb.JPG.af666644a7d912b8cd2c59b3afb5f38c.JPG

 

VGA Connector Board Mounted to 1088XLD Motherboard

VGA2GBS_mounted.thumb.JPG.26cf7295ab77b080d17781caa423e62a.JPG

 

RGB2VGA-XLD Board Fully Installed and Connected to VGA2GBS Connector Board

Overview.thumb.JPG.89e4919c3d15c99018f58f7dc68fd553.JPG

 

Rear Panel View Showing Standard VGA Connector

Rear_Panel_view.thumb.JPG.2d044c83b38571f529101dcf3d64bbcd.JPG

 

 

Dual Simultaneous Monitor Operation (RGB and VGA)

Dual_Monitors.thumb.JPG.8d9425f589a2b822eaf1102870f6347c.JPG

 

I still need to do some more tests, but so far all is looking very good.

 

And now the 1088XLD will have the same Video, Keyboard, and Mouse I/O as provided on the 1987 PS/2 computers. And when the video board is a VBXE the quality will be far superior. This arrangement would also work with a standard PS/2 KVM switch, thus sharing the peripherals between two similarly equipped 1088XLDs.

 

I figure by early December I'll be ready to release the design, including the revised GBSCTRL firmware.

 

  • Like 6
Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

Loading...
  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...